|
The Test - So how did I do the test? Basically the tests began at default settings. In this case 7
x 100 = 700 MHz. After that MDK2 Open GL demo and the 3d Mark 2000 did their thing. The same procedure was made for every time the FSB was increased. Some FSB levels were skipped between 100 FSB and 122 FSB. I did that because I presumed that the system would run stable at those speeds. It's actually the high speeds that are interesting. I also made notes about the stability of the system.
The Numbers:
|
FSB
|
MDK2
Demo
16/32Bit
|
3D MARK 2000
16/32Bit
|
STABILITY
|
| 100
(700 MHz) |
102,3 |
91,5 |
5856 |
4654 |
* |
| 115 (805 MHz) |
116,4 |
95,5 |
6490 |
4861 |
* |
| 122 (854 MHz) |
123,6 |
96,6 |
6760 |
4931 |
* |
| 126
(882 MHz) |
126,9 |
96,9 |
6873 |
5018 |
* |
| 133 (931 MHz) |
133,6 |
97,2 |
7076 |
5042 |
* |
| 138 (966 MHz) |
137,6 |
97,3 |
7261 |
5081 |
* |
| 140 (980 MHz) |
138,5 |
97,3 |
7289 |
5093 |
** |
| 142 (994 MHz) |
140,2 |
97,4 |
7332 |
5108 |
*** |
| 144 (1008 MHz) |
Application
hangs |
**** |
* System runs without any problems.
** Minor, intermittent problems such as Windows hang on shutdown, but nothing that affects the benchmark programs.
*** Intermittent problems with hang-ups during test runs. Problems disappeared when the volt core was increased to 1.7 volt.
**** The system starts up but hangs during test runs. Also tried to alter the core voltage without any success.
 The Conclusion - Well, it almost hit one GHz at 994 MHz. Close but it didn't reach all the way. MDK2 gained 6% from 100FSB to 142 FSB. With 3D Mark 2000 there is a 9 % gain of points. That doesn't seem like that much, but the FSB increased 42%. That was more than I dared hoping for.
Please take note that the settings were 32 bit in color in the textures. And the Z-buffer was set to 24 bit. With these settings at 16 bit I received a much higher score; 7369. And 140,9 with MDK2.
So the color depth is probably decisive in this case. Who knows, you might get over the magical (actually it's not that magical any longer) 1 GHz border with special cooling. But the temperature was never a problem during the test. Average CPU temperature was around 39 C/102 F. It rose a couple degrees during the nasty parts of the tests. Main reason for achieving these high FSB is foremost good components. Especially good memory and video card. But there is also an individual difference between CPU´s of same type. For an example I reached only 133 FSB with a PIII 700/100/256 SLOT1 CPU. And it did not have the same stability at 133 FSB as this one has at 142 FSB. So you have to be a little bit lucky when buying a Pentium III. But anyhow a PIII 700 is more valuable to over clockers than a PIII800. I managed to run a PIII 800 at 115 FSB = 920 MHz. And remember its $ 60 cheaper than the pIII800 that makes it even more worth its price.
Comments/ideas
?! Click here ...
|
